NCA-GENL Study Guide

This chapter builds the mental model you need for the rest of the book: what generative AI is solving, why modern LLMs are trained the way they are, and which deep learning principles directly affect quality, cost, and stability. If Chapter 1 is weak, every later chapter becomes memori

• - Generative vs Discriminative models
• - Supervised vs Unsupervised vs Self-supervised learning
• - Representation learning
• - Foundation models
• - Scaling laws

Transformers are the operating system of modern LLMs. This chapter explains how tokens move through attention blocks, why context length is expensive, and what model architecture choices imply for quality and serving cost.

• - Self-attention
• - Multi-head attention
• - Positional encoding
• - Encoder-only vs Decoder-only vs Encoder-Decoder
• - Feed-forward blocks

Training quality is a systems problem: objective design, data quality, optimi

• - Pretraining
• - Fine-tuning
• - Supervised fine-tuning (SFT)
• - Instruction tuning
• - Transfer learning

Prompt engineering is interface design for model behavior. In production, quality depends on both prompt structure and decoding policy. This chapter covers practical prompt patterns, inference controls, and security hardening.

• - Zero-shot
• - Few-shot
• - Chain-of-thought
• - Role prompting
• - System prompts

RAG connects LLMs to external knowledge so answers are grounded, current, and auditable. This chapter covers retrieval architecture, embedding strategy, chunking, ranking, and evaluation patterns that determine whether RAG helps or harms quality.

• - Retrieval-Augmented Generation (RAG)
• - Vector embeddings
• - Embedding models
• - Vector databases
• - Semantic search

Most teams cannot afford repeated full fine-tuning of large models. Parameter-efficient methods provide practical adaptation paths with lower cost and faster iteration. This chapter compares approaches and operational tradeoffs.

• - Full fine-tuning
• - PEFT
• - LoRA
• - Adapter layers
• - Prompt tuning

Alignment ensures models behave in ways users and organi

• - RLHF
• - Reward modeling
• - Human preference optimization
• - Policy optimization
• - Alignment techniques

Evaluation is how you convert "it seems good" into defensible release decisions. This chapter separates quality, performance, and reliability metrics, then shows how to combine them into practical release gates.

• - Perplexity
• - BLEU
• - ROUGE
• - F1 score
• - Exact match

LLM risk is not only model risk. It includes prompt channels, retrieval channels, tool calls, user interfaces, and governance processes. This chapter provides a layered approach to safety, security, and responsible AI operations.

• - Bias in LLMs
• - Fairness
• - Toxicity detection
• - Content filtering
• - Guardrails

Multimodal systems expand capability by linking text with image, audio, and video signals. They also expand risk, cost, and evaluation complexity. This chapter provides a practical understanding of major model families and deployment implications.

• - Multimodal models
• - Vision-language models
• - Text-to-image models
• - Diffusion models
• - GANs (high-level awareness)

Production success is determined by serving architecture, optimi

• - Quantization (INT8, FP16)
• - Pruning
• - Knowledge distillation
• - TensorRT optimization
• - Batch inference

Data workflows determine whether LLM systems stay reliable over time. This chapter covers ETL patterns, dataset/version governance, experiment tracking, monitoring, drift detection, and feedback loops needed for operational maturity.

• - Data pipelines
• - ETL workflows
• - Feature engineering
• - Embedding pipelines
• - Dataset labeling